Surveillance of Identifying Vehicles Parked In No Parking

Surveillance of Identifying Vehicles Parked In No Parking

Mentor:

Mr. Balmukund Mishra

Project Members:

Poonam Magadum

Aryan Songira

GitHub Link:  https://github.com/PoonamMagadum/Parking-Classification

Introduction:

Illegal parking has become a serious problem for many urban managers in large and growing cities. The resulting effect on vehicle movement can cause issues in traffic congestion, air-pollution and public safety. Traditionally, the detection of illegal parking is achieved through manual inspection, which requires considerable human efforts from law enforcement and security personals. In this work, we aim to reduce such effort by combining deep learning and computer vision.

With decreasing costs of high quality surveillance systems, human activity detection and tracking has become increasingly practical. Accordingly, automated systems have been designed for numerous detection tasks, but the task of detecting illegally parked vehicles has been left largely to the human operators of surveillance systems. We propose a methodology for detecting this event in real time. Unauthorized parking on city streets is a major contributor to traffic congestion and road accidents in developing nations. Vehicles parked in no parking area makes it difficult for vehicles to pass through the remaining road spam.

In this project, we present an algorithm for automated detection of illegally parked vehicles in real time. There were numerous methodologies for doing the vehicle detection and there's a ton progress in increasing the accuracy of the model from the basic model from extracting low level features from video caught from the reconnaissance cameras which can't cover the general view and passing them to classifiers like SVM (support Vector Machine), Naive Bayes, K-Nearest Neighbor and other classifiers to extracting every pixel from frames in videos that are captured from the drones which covers the whole view and passing them to the pre trained models like Imagenet, Resnet50, Vgg16.

Out of all the models for vehicle detection and recognition, the basic step in every vehicle detection model is the extraction of frames from every video and resizing all the frames to a equal size and passing them to the models, creating objects for the model and loading weights to the model and the final step is passing data for prediction where the model returns the class or label which is most similar to the input data of prediction. After an study of a great deal of models we built the 3D-CNN ( Convolutional Neural Network )

Project Description:

 

The aim of our project is to identify the vehicles parked in no parking zone.

Dataset Used:

 

The dataset that we have created contains videos that have been recorded with vehicles actions such as 'Running', 'Parked in parking zone', 'Parked in no parking zone'. These videos have been recorded at different times of day and from different angles to train model properly.

In the dataset every class consists of 200 videos to make the data balanced.

 

No.	Video Group	No. of Videos
1	Car Running	200
2	Parking	200
3	No Parking	200

Data Preprocessing:

 

 As we cannot directly pass a video to any model so we have to extract each frame from every video and  pass it to the model and we split the data into 2 parts that is training data and the testing data first we train the model with the training data and then test the model with the testing data and run the model with small changes in the parameters until we get a satisfied outcome.

Model:

 

3D-CNN (Three dimensional Convolutional Neural Network) model which we have created, trained and tested them for identifying vehicles parked in no parking area.

3D CNN:

 

We know that the 1D-CNN model is usually used for statistic knowledge and therefore the  2D-CNN model has the best accuracy in image classification and once it involves the 3D-CNN the dimension is for the statistic since videos area unit temporal knowledge. The output for the 3D-CNN is four dimensional. The 3D convolution is achieved by convolving a 3D kernel to the cube shaped by stacking multiple contiguous frames along. By this construction, the feature map within the convolution layer is connected to multiple contiguous frames.                                         

For the 3D-CNN model; we've resized each frame to 32,32,32 with depth thirty two and so we tend to passed the info to Conv3d layer , maxpooling 3d, flatten, dense, batch_normalization, dropout with activation functions  and extracted the utmost options from the input file and born-again it into the dimensions of labels or categories the dataset we've taken consists of three categories.

                                   

                                                  

                                                                 layers of 3d-cnn model

  

 The 3d-cnn architecture:

Graphs:

Accuracy of 3d-cnn model:

Loss of 3d-cnn model:

ACCURACY OBTAINED AFTER TRAINING MODEL WITH  3D-CNN METHOD IS   82%

Result:

We have considered 80% of data as training data and 20% of data as test data and we compiled the model with Adam optimizer.Adam was utilized as the advancement calculation since it is far superior to calculations like AdaGrad and RMSProp. This is on the grounds that Adam consolidates the best of their properties to manage loud issues. The measurements utilized for assessment was exactness.We got the highest accuracy with 3D-CNN model.

                                    

Conclusion and Future Work:

 

We have studied various models for usage of vehicle detection recognition and identify parked vehicles and out of those models we got the most elevated exactness for 3D-CNN. We are able to successfully detect illegally parked vehicles by accurately tracking all the vehicles in the scene. We can train the model on GPU's for more accuracy and deploy them in drones for identification of vehicles parked in no parking area.